Skip to content
Permalink
Branch: master
Commits on May 20, 2019
  1. compat: Update forward compatibility horizon to 2019-05-20

    tensorflower-gardener committed May 20, 2019
    PiperOrigin-RevId: 249007962
  2. Update GraphDef version to 41.

    tensorflower-gardener committed May 20, 2019
    PiperOrigin-RevId: 249007961
Commits on May 19, 2019
  1. Update GraphDef version to 40.

    tensorflower-gardener committed May 19, 2019
    PiperOrigin-RevId: 248922073
  2. compat: Update forward compatibility horizon to 2019-05-19

    tensorflower-gardener committed May 19, 2019
    PiperOrigin-RevId: 248922048
Commits on May 18, 2019
  1. Update GraphDef version to 39.

    tensorflower-gardener committed May 18, 2019
    PiperOrigin-RevId: 248849232
  2. compat: Update forward compatibility horizon to 2019-05-18

    tensorflower-gardener committed May 18, 2019
    PiperOrigin-RevId: 248849223
  3. Go: Update generated wrapper functions for TensorFlow ops.

    tensorflower-gardener committed May 18, 2019
    PiperOrigin-RevId: 248834474
  4. Fix bug in the docstring for spatial_softmax to say "ij" coordinate s…

    tensorflower-gardener committed May 18, 2019
    …ystem.
    
    PiperOrigin-RevId: 248819137
  5. Fix TPU strategy mirrored variable to directly use initialized value.

    tensorflower-gardener committed May 18, 2019
    PiperOrigin-RevId: 248813334
Commits on May 17, 2019
  1. Merge pull request #28780 from chengchingwen:peter/ckptcapi

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248792802
  2. Support quantization to float16

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248790891
  3. Merge pull request #28809 from ppwwyyxx:patch-18

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248789454
  4. Merge pull request #28483 from pranavm-nvidia:develop-pranavm-mm-only…

    tensorflower-gardener committed May 17, 2019
    …-changes
    
    PiperOrigin-RevId: 248788115
  5. Go: Update generated wrapper functions for TensorFlow ops.

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248786571
  6. Merge pull request #28771 from ROCmSoftwarePlatform:google-upstream-p…

    tensorflower-gardener committed May 17, 2019
    …r-update-rocprim
    
    PiperOrigin-RevId: 248776835
  7. Merge pull request #26896 from ROCmSoftwarePlatform:google_upstream_e…

    tensorflower-gardener committed May 17, 2019
    …xtract_volume_patches_op
    
    PiperOrigin-RevId: 248776319
  8. Merge pull request #28784 from Intel-tensorflow:mkl-fix-ubuntu18.04

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248774800
  9. Merge pull request #28571 from ROCmSoftwarePlatform:google-upstream-p…

    tensorflower-gardener committed May 17, 2019
    …r-cuda-device-alias
    
    PiperOrigin-RevId: 248770608
  10. Updates the TensorFlowLiteC podspec description.

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248768833
  11. Merge pull request #24018 from yongtang:21964-sparse.concat-shape

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248764459
  12. Go: Update generated wrapper functions for TensorFlow ops.

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248763861
  13. [TF:XLA] Fix tf2xla's in_topk lowering and update tests.

    tensorflower-gardener committed May 17, 2019
    - Fix an off by 1 comparison error.
    - Model Tensorflow's in_topk behavior for non-finite values.
    - Update Tensorflow's documentation to indicate behavior on non-finite values.
    - Enable TF:Classic's in_topk tests to run on XLA.
    - Correct test cases to check for off-by-1 errors and non-finite values.
    
    PiperOrigin-RevId: 248763768
  14. Merge pull request #27167 from houtoms:pr_batch_norm_ex_4

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248763766
  15. Implement _matvec for LinearOperatorDiag to avoid extra dimensions / …

    tensorflower-gardener committed May 17, 2019
    …squeezing.
    
    PiperOrigin-RevId: 248762369
  16. The C++ kernel of gather op supports batch dimensions.

    tensorflower-gardener committed May 17, 2019
    The C++ implementation of gather now takes a batch_dims argument and works in the same way as ResourceVariable's gather kernel.
    By default, batch_dims is set to 0, for compatibility with existing code.
    END_PUBLIC
    
    RELNOTES: The C++ kernel of gather op supports batch dimensions.
    
    BEGIN_PUBLIC
    The XLA kernel of Gather op supports batch dimensions.
    
    PiperOrigin-RevId: 248758666
  17. Merge pull request #23850 from howl-anderson:master

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248757421
  18. Adds the TFLite version string to the Swift library.

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248755089
  19. Merge pull request #28768 from bioothod:master

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248754577
  20. Merge pull request #28744 from samikama:GPUTestOutputFix

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248748011
  21. Go: Update generated wrapper functions for TensorFlow ops.

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248744820
  22. Removes min Bazel version check when building for Apple platforms.

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248742379
  23. Make shape relaxation opt-in in tf.function.

    tensorflower-gardener committed May 17, 2019
    By default shape relaxation will be off. Add a new optional experimental_relax_shapes argument (defaulting to False) to the tf.function decorator. Setting this to true enables shape relaxation for the decorated function. These same changes are also reflected in defun.
    
    Also fixing a bug that caused functions with an explicit input signature to always miss the cache.
    
    PiperOrigin-RevId: 248736376
  24. DepthwiseConv, dot-product kernel asm, add dot-product kernels to dis…

    tensorflower-gardener committed May 17, 2019
    …patch.
    
    PiperOrigin-RevId: 248730608
  25. Merge pull request #27167 from houtoms:pr_batch_norm_ex_4

    tensorflower-gardener committed May 17, 2019
    PiperOrigin-RevId: 248726654
  26. DepthwiseConv, dot-product kernel asm, enable code to call dot-produc…

    tensorflower-gardener committed May 17, 2019
    …t directly.
    
    PiperOrigin-RevId: 248723199
Older
You can’t perform that action at this time.